alumina catalyst for the efficient growth of vertically-aligned carbon nanotube forests

Materials Science & Engineering B 245 (2019) 37–46

Contents lists available at ScienceDirect

Materials Science & Engineering B journal homepage: www.elsevier.com/locate/mseb

One-pot preparation of iron/alumina catalyst for the efficient growth of vertically-aligned carbon nanotube forests

T

Arthur Rousseya,1, Nicolas Veniera, Hussein Fneichb, Lucas Giardellac, Thomas Pinaudb, ⁎ Saïd Tahirc, Mario Pelaez-Fernandezd, Raul Arenald,e,f, Ahmad Mehdib, Vincent Jourdainc, a

Global Technologies, 496 Avenue Francis Perrin, Rousset, France Institut Charles Gerhardt, UMR5253 CNRS, ENSCM, Université Montpellier, Place Eugène Bataillon, 34095 Montpellier Cedex 05, France c Laboratoire Charles Coulomb, CNRS, Univ. Montpellier, Montpellier, France d Laboratorio de Microscopias Avanzadas (LMA), Instituto de Nanociencia de Aragon (INA), Universidad de Zaragoza, C/Mariano Esquilor s/n, 50018 Zaragoza, Spain e ARAID Foundation, 50018 Zaragoza, Spain f Instituto de Ciencias de Materiales de Aragon, CSIC-U. de Zaragoza, Calle Pedro Cerbuna 12, 50009 Zaragoza, Spain b

A R T I C LE I N FO

A B S T R A C T

Keywords: Nanotubes Vertically-aligned carbon nanotubes Catalytic growth Fe(NO3)3 Al(NO3)3 Dip-coating CVD Water-assisted growth Silica Alumina Aluminum hydroxides

The catalytic growth of vertically-aligned carbon nanotubes (VA-CNTs) forest usually requires thin catalyst films deposited by multi-step and costly physical vapor deposition techniques. Here, we demonstrate that an efficient catalyst and its supporting layer for VACNT growth can be prepared by using a simple solution of Fe(NO3)3 and Al(NO3)3 deposited on silica in a single step. This process being much simpler and cheaper than existing preparation methods, it can easily be transferred to industry for the low-cost, thin and large-area coating of catalyst for VA-CNT growth. Our study shows that aluminum hydroxides preferentially react with the SiO2 surface while iron hydroxides tend to form oxide or hydroxide nanoparticles, thus allowing preparation of an aluminum-based buffer layer with iron-based nanoparticles at its surface. Optimization of the Fe/Al ratio and salt concentrations yielded catalysts with performances similar to standard Fe/Al2O3 catalysts prepared by physical vapor deposition.

1. Introduction During the last 25 years, carbon nanotubes (CNTs) have raised a great interest due to their unique structural and physical properties. Vertically aligned CNTs (VA-CNTs) arrays in particular have shown great potential for many applications such as field emission [1,2], energy storage [3,4], gas sensors [5,6], membranes [7,8], structural composites [9,10] or thermal interface [11,12]. Methods to produce tall and high-quality arrays of VA-CNTs are now well established, especially using the water-assisted catalytic chemical vapor deposition (C-CVD) process, the so-called “super-growth” [13]. Industrial manufacturing of VA-CNTs is on the rise but their production costs remain high, which strongly hinders the commercialization and wide application of VACNT-based materials. The growth process of VA-CNTs typically involves a carbon feedstock decomposed at high temperature on metal catalyst nanoparticles supported on an oxide layer, such as SiO2 [14], MgO [15], or Al2O3

[13,16], which acts as a buffer layer to prevent catalyst ripening and diffusion in the bulk of the support (usually a silicon wafer). To date, the best and standard catalyst system for growing dense and tall VACNTs is a thin Fe film (0.4–2 nm) supported on an Al2O3 underlayer (10–100 nm) [17]. Al2O3 is particularly efficient as its role is not limited to a simple diffusion barrier but is also believed to reduce hydrocarbon contamination of the surface in the presence of H2O [18] and to stabilize the oxidation state of iron nanoparticles (Fe2+ and Fe3+) [19], restricting iron mobility on the surface, and therefore nanoparticle sintering [20]. In most studies to date on VA-CNT growth, the buffer and catalyst layers are typically prepared by physical vapor deposition (PVD) [21]. Although PVD systems are widely used in the semiconductor industry, a less expensive and demanding process of catalyst deposition would be highly beneficial for the large-scale and continuous production of VA-CNT arrays [22]. Our work was therefore motivated by the need for a simpler and cheaper method of catalyst preparation for the large-scale industrial production of VA-CNTs. Methods of wet

⁎

Corresponding author. E-mail address: [email protected] (V. Jourdain). 1 Present address: Grenoble Alpes University, CEA-LITEN, 17 rue des Martyrs, 38054 Grenoble Cedex 09, France. https://doi.org/10.1016/j.mseb.2019.05.005 Received 19 August 2018; Received in revised form 27 March 2019; Accepted 6 May 2019 Available online 13 May 2019 0921-5107/ © 2019 Elsevier B.V. All rights reserved.

Materials Science & Engineering B 245 (2019) 37–46

A. Roussey, et al.

temperature to 750 °C. At 750 °C, 200 ppm of H2O were added using a small flow of He in a water bubbler which was maintained at low temperature (typically 5 °C) and the samples were kept at 750 °C in this atmosphere for 5 min. Humidity was controlled using an hygrometer (MIS Probe 2, General Electrics) prior to each experiment. Then, the samples were exposed for 10 min to a He/C2H4 mixture (810/190 sccm) with 200 ppm H2O to grow VA-CNTs. The furnace was then cooled under He at a temperature below 100 °C before opening and withdrawal of the samples to prevent VACNT oxidation. The samples presented in this study were prepared in four different runs. During each run, a reference catalyst prepared by PVD with 1 nm of Fe on 20 nm of Al2O3 (PVD: Fe@Al2O3) was added in the furnace and the thickness of VACNT grown on it was controlled by an optical camera to confirm the reproducibility of the growth conditions independently of the environmental conditions.

deposition of metal (Fe, Co, Ni,…) have already been reported on alumina or silica, using metal salt solutions or metal colloid suspensions as starting materials. The as-made catalysts showed activities comparable to those of PVD-prepared catalysts [22–26]. Wet-deposition methods were also developed to prepare the Al2O3 underlayer [27–29]. For example, Wang et al. developed a fully wet procedure using boehmite nanoplates (γ-AlO(OH)) deposited on a silicon chip, which were converted in a 20-nm thick Al2O3 buffer layer by annealing at 750 °C in air. After deposition of a colloidal suspension of Fe3O4 nanoparticles, they obtained a catalyst yielding millimeter-thick VA-CNT arrays [30]. However, this approach remains complex and requires several steps: i) preparation and purification of the boehmite solution, ii) deposition of the particles, iii) annealing to form an Al2O3 layer, iv) preparation and deposition of the Fe2O3 colloidal suspension. Our goal was therefore to build on the versatility of wet methods while developing a simpler and cheaper process than those previously reported. Here, we report a simple and fully wet approach to prepare catalysts able to grow dense and tall VA-CNT arrays on oxidized silicon wafers. The main novelty is that the aluminum-based buffer layer and the catalyst nanoparticles at its surface are formed together in a single step. This method is based on the dip-coating of a single solution of a mixture of Fe(NO3)3, Al(NO3)3 and NH4OH which are widely available and lowcost precursors. Although similar approaches were already tested in previous works, they did not yield tall VA-CNT arrays (< 50 µm) [31–33]. We show here that a careful optimization of the concentrations of the different species in the solution leads to a growth activity and a VA–CNT quality comparable to those obtained with typical PVDmade catalysts in the same growth conditions.

2.4. Characterization Scanning electron micrographs were obtained using a Hitachi S4800 operating at 10 kV. Raman characterization of the VA-CNTs was performed at a laser wavelength of 532 nm using a Horiba Jobin-Yvon T64000 spectrometer equipped with a silicon camera cooled with liquid nitrogen and a microscope. The focused laser spot on the sample was typically 1 µm. Atomic force micrographs were obtained using a Brucker AFM D3100 in tapping mode using a NCH point probe with a tip radius of 5 nm at F = 242.9 kHz, Q = 460 and k = 18 N·m−1. X-Ray Photoelectron spectroscopy (XPS) measurements were performed on Mepitel®. The excitation source produced a monochromatic Al Kα line (1486.6 eV) and worked at 100 W of power. The analyzed surface had a 400-µm diameter. All XPS components were assigned from the analysis of the values reported for reference compounds and referenced in the NIST database [34]. X-Ray diffraction (XRD) patterns were obtained using a PAN analytical X’Pert Pro MPD diffractometer at Cu Kα irradiation (λ = 1.5418 Å) with a step size of 0.033° (2θ scale) in the 12°−80° interval. Note that in the case of powders submitted to a reducing treatment for subsequent XRD characterization, the samples were stored under inert gas just after the reducing treatment to prevent oxidation in air. For other samples, they were normally exposed to air as during the elaboration process before XRD and oxidation in air was taken into account for the XRD interpretation. Transmission Electron Microscopy (TEM) was first performed on a JEOL 1200 EX II operated at 100 kV. High-Resolution TEM (HRTEM) micrographs were obtained with a FEI Titan Cs image aberration-corrected microscope working at 80 kV. In the latter case, the observations were performed at low temperature (approx. 77 K) to avoid electron irradiation damage of the nanotubes and possible contamination of the samples. Micrographs were subsequently analyzed using the Gatan Digital Micrograph software taking intensity profiles through each tube to measure their diameter distribution.

2. Experimental 2.1. Materials Fe(NO3)3·9H2O (ACS Reagent, > 98 %), Al(NO3)3·9H2O (ACS Reagent, > 98 %), NH4OH (5 M) were purchased from Sigma-Aldrich and used without further purification. He (99.995 %), H2 (99.9995 %) and C2H4 (99.95 %) were purchased from Linde Gas. Gas flow were controlled using Brooks GF80 mass flow controllers. Silicon wafers were thermally oxidized on both sides to reach a layer of 600 nm of SiO2. For reference, PVD Al2O3 sublayers were prepared by depositing 20 nm of Al2O3 by radio-frequency non-reactive sputtering. For control experiments with PVD Fe, 1 nm of Fe was further evaporated on top of the Al2O3 layer. Just prior to dip-coating, thermally oxidized Si substrates (ca. 15x50 mm2) were washed by immersion into an active NH4OH (5 M)/H2O2 (5 M)/H2O mixture for 15 min and then thoroughly washed with H2O. Substrates coated with Al2O3 were thoroughly washed with acetone, isopropanol and H2O. 2.2. Solution and catalyst preparation

3. Results

0.50 g of Fe(NO3)3·9H2O was added to a 100 mL solution of 0.025 M NH4OH to yield a clear brown solution (Fe solution, Fig. 1). Then, 0.46 g Al(NO3)3·9H2O was added to yield a transparent brown and acidic solution (pH ∼ 4) hereafter called FeAl[X] solution with X being the Fe/Al molar ratio in the solution. These solutions were then dipcoated at 20 mm·min−1 in a humidity- and temperature-controlled chamber (RH ≈ 50 % at 27 °C) on freshly cleaned thermally oxidized silicon wafers. The samples were dried at room temperature for 1–24 h before CVD experiments.

We first studied the influence of the catalyst preparation parameters on the features of the grown CNTs. Fig. 1 shows the typical procedure for the experiments conducted throughout this study. First, we investigated the activity of a solution of Fe(NO3)3 in dilute NH4OH (Fe solution). The role of NH4OH is to allow the formation of metal hydroxides able to condense into nanoparticles. This solution was dipcoated on a thermally oxidized Si wafer (Fe@SiO2) and on a silicon wafer coated with an extra layer of 20 nm of Al2O3 deposited by PVD (Fe@Al2O3). The activity of these catalysts was compared to that of a reference catalyst prepared by PVD with 1 nm of Fe on 20 nm of Al2O3 (PVD: Fe@Al2O3). In each run, the thickness of the VA-CNTs grown on the reference PVD catalyst was measured to be 660 µm ± 10 %, thus confirming the good reproducibility of the growth conditions. We observed that thick VA-CNT arrays were grown on both PVD:Fe@Al2O3

2.3. VA-CNT growth VA-CNT growth was performed using a standard water-assisted protocol [13]. Briefly, the samples were placed on top of a silicon sample holder (40 × 50 mm2) in a 2-inches tubular furnace and heated in a He/H2 mixture (600/400 sccm) during 15 min from room 38

Materials Science & Engineering B 245 (2019) 37–46

A. Roussey, et al.

Fig. 1. Protocols of catalyst preparation and corresponding catalyst nomenclature. Table 1 VA-CNT height, Raman G-band positions and G/D ratio (λ = 532 nm) of the samples grown from the different catalysts. Entry

Catalyst

[Fe] (mol·L−1)

[Fe]/[Al]

VA-CNT height (µm)

G band (cm−1)

G/D ratio

1 2 3 4 5 6 7 8 9 10 11 12

PVD: Fe@Al2O3 Fe@Al2O3 FeAl[1]@Al2O3 FeAl[0.2]@SiO2 FeAl[0.6]@SiO2 FeAl[0.8]@SiO2 FeAl[1]@SiO2 FeAl[1.5]@SiO2 FeAl[1]@SiO2 2xFe@Al2O3 2xFeAl[0.8]@SiO2 2xFeAl[1]@SiO2

N/A 0.0125 0.0125 0.0125 0.0125 0.0125 0.0125 0.0125 0.0125 0.25 0.25 0.25

N/A N/A 1 0.2 0.6 0.8 1 1.5 2 N/A 0.8 1

660 ± 10 % 590 690 420 660 550 550 450 250 620 n.d. 640

1584 1583 1581 1576 1576 1578 1579 1584 1577 1575 1580 1580

3.2 2.2 3.7 1.1 1.2 2.8 1.5 1.5 1.2 4.8 2.5 3.1

and Fe@Al2O3, with thickness of, respectively, 650 µm and 590 µm (Table 1, entries 1 and 2). At the opposite, a thin layer of entangled CNTs was grown with Fe@SiO2, in good agreement with previous studies performed in the absence of an alumina under-layer [35,36]. Second, we studied the influence of a PVD Al2O3 under-layer on the activity of the FeAl[X] catalysts. As shown in Table 1 (entry 3), a FeAl [1] solution dip-coated on PVD Al2O3 (FeAl[1]@Al2O3) yielded a VACNT array with a thickness of 690 µm, while a slightly thinner array (550 µm) was obtained with the same solution deposited on Si/SiO2 (FeAl[1]@SiO2, entry 7). Together, these results confirm the welldocumented effect that an Al2O3 under-layer strongly promotes VACNT growth from Fe catalyst nanoparticles. Third, we studied the influence of the Fe/Al ratio (X = [Fe]/[Al]) on the activity of the FeAl[X]@Al2O3 catalysts. The maximum VA-CNT thickness (660 µm) was obtained with a Fe/Al ratio of 0.6 (FeAl[0.6]@ SiO2, Table 1, entry 5). Importantly, this thickness is comparable to the one obtained with the reference PVD catalyst thus demonstrating that catalysts prepared by a simple dip-coating step can yield VA-CNT forests with heights similar to PVD-made catalysts. Higher and lower Fe/

Al ratio yielded thinner VA-CNT layers (Table 1, entries 4–9). Fourth, the effect of increasing the total concentration of Fe and Al salts was investigated. By doubling the total concentration in Fe and Al salts, significantly taller VA-CNTs were obtained: 640 µm for 2xFeAl[1] @SiO2 (entry 12) by comparison to 550 µm for FeAl[1]@SiO2 (entry 7). Note that for Fe solution deposited on PVD Al2O3, doubling the Fe concentration has little effect: 620 µm for 2xFe@Al2O3 (entry 10) to be compared to 590 µm for Fe@Al2O3 (entry 2). All together, these results highlight the high potential of this fully-wet single-step approach since a simple coating by Fe and Al salt solutions directly on Si/SiO2 shows performances comparable to those of PVD catalysts. The morphologies of the VA-CNT arrays were then analyzed by scanning electron microscopy (SEM) as shown in Fig. 2. SEM pictures showed well aligned CNTs on all the samples grown on Al2O3 (Fig. 2a–c) while less aligned and curly CNTs were observed for the samples grown from FeAl[X]@SiO2 (Fig. 2d–h) with X between 0.2 and 1.5. Since alignment is essentially caused by steric interactions between CNTs, this supports that FeAl[X]@SiO2 catalysts yield a lower CNT density of the VACNT forests than PVD catalysts. In general, a 39

Materials Science & Engineering B 245 (2019) 37–46

A. Roussey, et al.

Fig. 2. SEM pictures of the VA-CNTs grown from: a. PVD: Fe@Al2O3, b. Fe@Al2O3, c. FeAl[1]@Al2O3, d. Fe@SiO2, e. FeAl[0.2]@SiO2, f. FeAl[0.6]@SiO2, g. FeAl [0.8]@SiO2, h. FeAl[1.5]@SiO2, i. 2xFe@ Al2O3, j. 2xFeAl[1]@SiO2.

280 cm−1) associated with the radial breathing modes (RBMs) of small diameter CNTs (d = 1–3 nm). The occurrence of RBMs is systematically associated with higher intensities of the G and 2D bands and to downshifted D and 2D bands as also expected when moving toward such smaller-diameter CNTs. Since CNTs of 1–3 nm were not observed during our HRTEM observations but display an intrinsically higher Raman cross section due to the resonance effect, the results agree with samples mainly composed of double/triple-wall CNTs with 3–7 nm diameter with traces of CNTs with 1–3 nm diameter. The proportion of small-diameter CNTs was generally found higher for the catalysts deposited on PVD alumina. Note that having a mixture of large-diameter CNTs (weakly resonant and with low G/D ratio) and small-diameter CNT (highly resonant and with high G/D ratio) hinders the use of the G/D ratio (Table 1) to evaluate the overall crystalline purity of the sample. Instead, for the samples studied here, a higher G/D ratio essentially denotes a higher proportion of small-diameter CNTs in agreement with the other Raman features (more intense RBMs, downshifted D and 2D bands). By comparing the Raman spectra at different positions, one may roughly estimate that large- and small-diameter CNTs display a G/D ratio in the range of 1–2 and 2–6, respectively and quite independently of the catalyst. Of course, this value is expected to strongly vary with the growth conditions (temperature, precursor pressure) used [37]. Overall, the TEM and Raman results shows that our wet-based catalyst preparation method allows the preparation of VACNT arrays with similar morphology and structural quality than

lower homogeneity in CNT height was observed in VA-CNTs grown from FeAl[X]@SiO2 catalysts when compared to PVD catalysts, which also points toward a less homogeneous and less dense distribution of active catalyst particles. The diameter distribution of the CNTs grown from PVD: Fe@Al2O3, Fe@Al2O3, and FeAl[1]@SiO2 were evaluated by TEM and found to be comparable for the three samples with most tubes having diameters of 5 ± 1 nm. The structure of the CNTs grown from FeAl[1]@SiO2 was further analyzed by high-resolution TEM showing that mainly doubleand triple-wall CNTs were obtained (Fig. 3, mean size 5.6 ± 1.6 nm with a ratio of 2.4 DWCNTs for 1 TWCNT). By comparison, VA-CNTs grown from Fe@Al2O3 showed slightly thinner nanotubes with a higher proportion of DWCNTs (5.2 ± 0.8 nm with 3.3 DWCNTs for 1 TWCNT, Fig. 3). The crystalline quality of all CNTs was found relatively good and comparable to that reported in the literature for VA-CNTs grown from PVD-made catalysts. The samples were then analyzed by micro-Raman spectroscopy (Fig. 4). All the VA-CNTs samples showed a G band at around 1580 cm−1 (see Table 1) as expected from a sample dominated by multi-walled CNTs (MWCNTs) with diameters in the range of 3–7 nm as measured by HRTEM (Fig. 3). All samples also displayed D and 2D (G′) bands at positions in the range of 1327–1338 cm−1 and 2642–2670 cm−1, respectively. This is also in good agreement with the positions expected for MWCNTs at 532 nm of laser excitation. Many samples showed additional signals at low frequencies (less than 40

Materials Science & Engineering B 245 (2019) 37–46

A. Roussey, et al.

Fig. 3. a, b) Representative HRTEM pictures of CNTs and catalyst crystalline nanoparticles (highlighted by a red contour) after growth using the FeAl[1]@SiO2 catalyst. In b), a Fast Fourier Transform (FFT) has been applied to the TEM image and the crystalline spots corresponding to the catalyst NPs have been selected to apply an inverse FFT for building the insert image at the top left. c, d) Outer tubes diameter distribution diagram of FeAl[1]@SiO2 (c) and Fe@Al2O3 (d), respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

in good agreement with our SEM observations of less dense and curvy VA-CNTs [39] obtained with the catalysts prepared by dip-coating. From AFM measurements, it is clear that a severe surface reconstruction occurs during the H2 treatment. To further investigate the composition of the thin catalyst film before CVD experiments, XRD measurements of the solid residue obtained after evaporation of the different solutions were performed. Before reduction, the solid residue of a Fe(NO3)3 aqueous solution with 1 equivalent of NH4OH (Fe solution) showed the presence of crystalline NH4NO3 (Fig. 6a) with average crystal sizes > 100 nm. The same result was obtained for the solid residue of a Al(NO3)3 aqueous solution with 1 equivalent of NH4OH (Fig. 6b). The residue of FeAl[1] also showed the presence of crystalline NH4NO3 nanoparticles, along with crystalline FeO nanoparticles (Fig. 6c). However, there was no signature of crystalline Al-based compounds. Interestingly, when solution FeAl[1] was coated on high surface-area silica and reduced under H2, no diffraction peaks were observed, which indicates that large metallic Fe nanoparticles were not formed in agreement with AFM observations (Fig. 6d). The disappearance of the large NH4NO3 crystal nanoparticles previously observed provides an explanation for the evolution of surface roughness observed by AFM before and after H2 reduction as further supported by the decomposition temperature of NH4NO3 (210 °C).

standard PVD catalyst, yet with the use of simpler protocol and laboratory equipment.

4. Discussion To understand the effect of the addition of the Al salt, the catalyst formation mechanism was investigated using Atomic Force Microscopy (AFM), X-ray Diffraction (XRD) and X-ray Photoelectron Spectroscopy (XPS). AFM observations of the samples after deposition of solution FeAl [1] on Si/SiO2 and drying were first performed. As shown in Fig. 5a, large nanoparticle aggregates of 10–50 nm were observed on the surface. Since the diameter of a MWCNT is strongly related to the size of its catalyst particle [36,38], these particles are not likely to be directly responsible for the observed catalytic activity since the CNT diameter distribution was determined to be 5.6 ± 1.6 nm (Fig. 3c). Interestingly, after annealing at 750 °C under a He/H2 mixture with 200 ppm H2O and rapid cooling to RT under a He/H2 flow, well-defined and smaller nanoparticles were observed (height around 5 nm) on the surface (Fig. 5b). When compared to the PVD catalyst (Fig. 5c) whose surface is very homogeneous after reduction, FeAl[1]@SiO2 surface shows the presence of some larger nanoparticles (around 10 nm in height). This is 41

Materials Science & Engineering B 245 (2019) 37–46

A. Roussey, et al.

not continuous. The Fe 2p3/2 profile is made of two contributions: an Fe (II) contribution at 709.8 eV and an Fe(III) contribution at 711.1 eV with a Fe(II)/Fe(III) ratio of 1.65, plus two satellite peaks at higher energy (Fig. 7a). This excludes the presence of metallic iron (expected at ∼707 eV), of iron silicide (Fe3Si expected at 707.5 eV) and of iron silicate (Fe2SiO4 expected at 708.9–709.0 eV). It also excludes the possibility of a single Fe(II) or Fe(III) compound, or of Fe3O4 only (which has an Fe(II)/Fe(III) ratio of 0.5). For the Fe(III) peak at 711.1 eV, the most likely assignment is FeO(OH) (expected at 711.0–711.8 eV) or alternatively Fe2O3 (which is expected at 710.7–711.4 eV). For the Fe(II) peak at 709.8 eV, the most likely assignment is FeO which is expected at 709.6–710.3 eV. The Al 2p peak is found at 74.8 eV (Fig. 7c) which would nicely agree with an aluminosilicate such as Al2OSiO4 (expected at 74.5–74.9 eV) or an aluminum silicate hydroxide such as Al2Si4O10(OH)2 (expected at ∼74.7 eV), which in both cases would support a strong anchoring of aluminum with the SiO2 substrate. After H2 annealing, an even larger atomic contribution of Si (36.6 %) was observed, with 3.7 % of Al and 1.2 % of Fe remaining on the surface (Table 2), showing significant Fe diffusion in the SiO2 matrix at high temperature (750 °C). The Fe 2p3/2 profile displayed little change with still two contributions: an Fe(II) contribution at 710.1 eV and an Fe(III) contribution at 711.4 eV with about the same Fe(II)/Fe (III) ratio as before reduction (Fig. 7b). At the opposite, the Al 2p peak was strongly downshifted to 74.0 eV (Fig. 7c), which could be assigned to Al(OH)3 (expected at 73.9–74.4 eV) or alternatively to FeAl2O4 (expected at ∼74.3 eV). The latter one should give rise to an Fe 2p3/2 contribution at ∼710.0 eV which is in good agreement with the Fe(II) peak observed at 710.1 eV. We performed the same H2 treatment and XPS analysis on the same solution without Al(NO4)3 (i.e. Fe@SiO2) and found that the amount of Fe remaining at the surface was about three times less (Table 2). This provides an additional support for the formation of an iron aluminate such as FeAl2O4 which would stabilize Fe at the surface of SiO2. The remaining Fe(III) peak is assigned to Fe2O3 and/or FeO(OH) formed from reduced iron when exposed to air (Fig. 7d). Taken together, these results support the following picture (Fig. 8). During air drying, Al(OH)3 tends to react with the SiO2 surface (e.g. with hydroxyl groups) to form a surface layer of aluminosilicate or aluminum silicate hydroxide compounds, while Fe(OH)3 tends to form FeO or FeO(OH) nanoparticles. During H2 annealing, particles of Fe oxides and hydroxides can either get reduced to metallic iron which then tends to diffuse into the SiO2 substrate. Alternatively, in the presence of surface aluminum oxides or hydroxides, they can form surface layers of iron aluminates which are less reducible and act as a buffer layer which stabilizes small iron nanoparticles at their surface. So, beside its standard role of diffusion barrier, SiO2 plays another role in our process, by allowing the formation of a mixed Al-Si oxidized layer which will anchor and stabilize catalyst nanoparticles. If exposed to a carbon source during CVD, these stabilized iron particles will allow the growth of long CNTs. Instead, if exposed to air, these iron nanoparticles will quickly become oxidized into Fe(III) compounds such as Fe2O3 or FeO(OH). This mechanism supports that increasing the concentration of Al in the solution at constant ratio with Fe should increase the surface coverage with aluminum compounds and therefore a higher density of stabilized iron particles and so a higher yield of long and defective nanotubes: this is in perfect line with our previous observation that doubling the proportion of Fe and Al promotes both a taller VACNT forest and a lower defect density (see entry 12 of table 1).

Fig. 4. Raman spectra of VA-CNT arrays grown from the different catalysts.

XPS analyses were then performed. Surface analysis of sample FeAl [1]@SiO2 after drying but before reduction under H2, showed a very large contribution of the SiO2 substrate with 25.5 % of Si, and only 5.0 % of Al and 5.0 % of Fe (Table 2). Since XPS is only sensitive to the top 10 nm of the surface, this indicates that the deposited layer is very thin (less than 10 nm) and/or

5. Conclusion In this study, a fully wet process was developed to prepare, in a single deposition step, a catalyst to grow VA-CNT arrays on standard SiO2/Si wafers. This is important from a materials engineering point of view because the process is much simpler and cheaper than existing 42

Materials Science & Engineering B 245 (2019) 37–46

A. Roussey, et al.

Fig. 5. AFM images of a. FeAl[1]@SiO2 before annealing, b. FeAl[1]@SiO2 after annealing at 750 °C under H2 and c. PVD: Fe@Al2O3 after annealing at 750 °C under H2 .

Fig. 6. a. XRD of the dry residue of Fe(NO3)3 aqueous solution with 1 equivalent of NH4OH before reduction showing mainly the presence of NH4NO3 [40] (ICSD collection code 002772). b. XRD of the dry residue of a Al(NO3)3 aqueous solution with 1 equivalent of NH4OH before reduction showing mainly the presence of NH4NO3 [40] c. XRD of a dry residue of a FeAl[1] solution showing mainly the presence of Wuestite (Iron Oxide Fe0.902O) [41] (ICSD collection code 040089) and NH4NO3 [40]. d. XRD of FeAl[1] impregnated on high surface area silica solution after reduction under H2 at 750 °C showing only the typical large shoulder of amorphous silica.

preparation methods. We therefore expect it to be easily transferred to industry for low-cost and large-area coating of catalyst for VA-CNT growth. This catalyst, prepared from a mixture of Fe(NO3)3 and an Al (NO3)3, yields VACNT height comparable to those of Fe/Al2O3 catalysts prepared by PVD which is standardly used for VA-CNT growth. Our study showed that, when mixed together, aluminum hydroxides preferentially react with the SiO2 surface while iron hydroxides tend to react together to form oxide or hydroxide nanoparticles: this explains

Table 2 Surface composition of the catalyst: elemental composition from XPS. Sample Name

Treatment

Si (2p)

O (1s)

Fe (2p)

Al (2p)

FeAl[1]@SiO2 FeAl[1]@SiO2 Fe@SiO2 Fe@Al2O3

After After After After

25.7 36.7 41.1 0.0

64.2 58.3 58.4 55.8

5.1 1.2 0.5 1.0

5.1 3.7 0.0 43.1

deposition H2 reduction H2 reduction H2 reduction

43

Materials Science & Engineering B 245 (2019) 37–46

A. Roussey, et al.

Fig. 7. Fe 2p XPS spectra of a) FeAl[1]@SiO2 before H2 reduction, and b) after H2 reduction; c) Al 2p XPS spectra of FeAl[1]@SiO2 before H2 reduction (blue), and after H2 reduction (black) compared to Fe@Al2O3 after H2 reduction (red); Fe 2p XPS spectra of d) Fe@SiO2 after H2 reduction and e) Fe@Al2O3 after H2 reduction. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 8. Proposed mechanism for catalyst formation and evolution.

44

Materials Science & Engineering B 245 (2019) 37–46

A. Roussey, et al.

why a single solution step can be used to prepare both an aluminumbased buffer layer and catalyst nanoparticles at its surface. This is important from a materials science point of view because it shows that chemical affinities between reactants can be played with to elaborate the catalyst nanoparticles and their supporting layer together in a single step. Following this initial demonstration, the approach may be further optimized and generalized to other catalysts to increase the homogeneity and density of the CNT forest and obtain a better control of the CNT diameter and crystalline quality. This simple and versatile approach appears as a strong alternative to PVD to easily prepare VA-CNT forests on various substrates, such as curved, porous or conducting substrates. Importantly for applications in electronics and electrochemistry requiring to electrically contact VA-CNT arrays, this process requires only a very thin layer of insulating oxide material, therefore allowing a more intimate contact between CNTs and their substrate.

[12] J. Liu, T. Wang, Y. Fu, L. Ye, Use of carbon nanotubes in potential electronics packaging applications, in 10th IEEE International Conference on Nanotechnology (IEEE-NANO), Seoul, South Korea, IEEE, 2010, 160-166. https://doi.org/10.1109/ NANO.2010.5698052. [13] K. Hata, D.N. Futaba, K. Mizuno, T. Namai, M. Yumura, S. Iijima, Water-assisted highly efficient synthesis of impurity-free single-walled carbon nanotubes, Science 306 (2004) 1362–1364, https://doi.org/10.1126/science.1104962. [14] S. Maruyama, E. Einarsson, Y. Murakami, T. Edamura, Growth process of vertically aligned single-walled carbon nanotubes, Chem. Phys. Lett. 403 (2005) 320–323, https://doi.org/10.1016/j.cplett.2005.01.031. [15] G.-Y. Xiong, D.Z. Wang, Z.F. Ren, Aligned millimeter-long carbon nanotube arrays grown on single crystal magnesia, Carbon 44 (2006) 969–973, https://doi.org/10. 1016/j.carbon.2005.10.015. [16] S. Esconjauregui, M. Fouquet, B.C. Bayer, C. Ducati, R. Smajda, S. Hofmann, J. Robertson, Growth of ultrahigh density vertically aligned carbon nanotube forests for interconnects, ACS Nano 4 (2010) 7431–7436, https://doi.org/10.1021/ nn1025675. [17] S.P. Patole, P.S. Alegaonkar, H.-C. Lee, J.-B. Yoo, Optimization of water assisted chemical vapor deposition parameters for super growth of carbon nanotubes, Carbon 46 (2008) 1987–1993, https://doi.org/10.1016/j.carbon.2008.08.009. [18] N. Suguru, H. Kei, S. Hisashi, K. Kazunori, Z. Zhengyi, M. Shigeo, Y. Yukio, Millimeter-thick single-walled carbon nanotube forests: hidden role of catalyst support, Jpn. J. Appl. Phys. 46 (2007) L399, https://doi.org/10.1143/JJAP.46. L399. [19] C. Mattevi, C.T. Wirth, S. Hofmann, R. Blume, M. Cantoro, C. Ducati, C. Cepek, A. Knop-Gericke, S. Milne, C. Castellarin-Cudia, S. Dolafi, A. Goldoni, R. Schloegl, J. Robertson, In-situ X-ray photoelectron spectroscopy study of catalyst−support interactions and growth of carbon nanotube forests, J. Phys. Chem. C 112 (2008) 12207–12213, https://doi.org/10.1021/jp802474g. [20] P.B. Amama, C.L. Pint, S.M. Kim, L. McJilton, K.G. Eyink, E.A. Stach, R.H. Hauge, B. Maruyama, Influence of alumina type on the evolution and activity of aluminasupported Fe catalysts in single-walled carbon nanotube carpet growth, ACS Nano 4 (2010) 895–904, https://doi.org/10.1021/nn901700u. [21] S. Yasuda, D.N. Futaba, T. Yamada, J. Satou, A. Shibuya, H. Takai, K. Arakawa, M. Yumura, K. Hata, Improved and large area single-walled carbon nanotube forest growth by controlling the gas flow direction, ACS Nano 3 (2009) 4164–4170, https://doi.org/10.1021/nn9007302. [22] S. Sakurai, H. Nishino, D.N. Futaba, S. Yasuda, T. Yamada, A. Maigne, Y. Matsuo, E. Nakamura, M. Yumura, K. Hata, Role of subsurface diffusion and ostwald ripening in catalyst formation for single-walled carbon nanotube forest growth, J. Am. Chem. Soc. 134 (2012) 2148–2153, https://doi.org/10.1021/ja208706c. [23] J.D. Beard, J. Stringer, O.R. Ghita, P.J. Smith, High yield growth of patterned vertically aligned carbon nanotubes using inkjet-printed catalyst, ACS Appl. Mater. Interfaces 5 (2013) 9785–9790, https://doi.org/10.1021/am402942q. [24] K. Mizuno, K. Hata, T. Saito, S. Ohshima, M. Yumura, S. Iijima, Selective matching of catalyst element and carbon source in single-walled carbon nanotube synthesis on silicon substrates, J. Phys. Chem. B 109 (2005) 2632–2637, https://doi.org/10. 1021/jp0454117. [25] H. Nishino, S. Yasuda, T. Namai, D.N. Futaba, T. Yamada, M. Yumura, S. Iijima, K. Hata, Water-assisted highly efficient synthesis of single-walled carbon nanotubes forests from colloidal nanoparticle catalysts, J. Phys. Chem. C 111 (2007) 17961–17965, https://doi.org/10.1021/jp0723719. [26] E.S. Polsen, M. Bedewy, A.J. Hart, Decoupled control of carbon nanotube forest density and diameter by continuous-feed convective assembly of catalyst particles, Small 9 (2013) 2564–2575, https://doi.org/10.1002/smll.201202878. [27] S. Dörfler, A. Meier, S. Thieme, P. Németh, H. Althues, S. Kaskel, Wet-chemical catalyst deposition for scalable synthesis of vertical aligned carbon nanotubes on metal substrates, Chem. Phys. Lett. 511 (2011) 288–293, https://doi.org/10.1016/ j.cplett.2011.06.027. [28] H. Wang, C. Na, Chemical bath deposition of aluminum oxide buffer on curved surfaces for growing aligned carbon nanotube arrays, Langmuir 31 (2015) 7401–7409, https://doi.org/10.1021/acs.langmuir.5b01002. [29] N.T. Alvarez, C.E. Hamilton, C.L. Pint, A. Orbaek, J. Yao, A.L. Frosinini, A.R. Barron, J.M. Tour, R.H. Hauge, Wet catalyst-support films for production of vertically aligned carbon nanotubes, ACS Appl. Mater. Interfaces 2 (2010) 1851–1856, https://doi.org/10.1021/am100128m. [30] H. Wang, C. Na, Synthesis of millimeter-long vertically aligned carbon nanotube arrays on aluminum oxide buffer prepared by layer-by-layer assembly of boehmite nanoplates, Carbon 66 (2014) 727–729, https://doi.org/10.1016/j.carbon.2013. 09.034. [31] J.-M. Bonard, P. Chauvin, C. Klinke, Monodisperse multiwall carbon nanotubes obtained with ferritin as catalyst, Nano Lett. 2 (2002) 665–667, https://doi.org/10. 1021/nl0255606. [32] L.Q. Fei L, Q.X. Jia, H.M. Luo, Polymer-assisted deposition of composite catalysts for the growth of vertical aligned carbon nanotubes, Chem. Eng. Process Technol. 1 (2013) 1013. [33] P.M. Parthangal, R.E. Cavicchi, M.R. Zachariah, A generic process of growing aligned carbon nanotube arrays on metals and metal alloys, Nanotechnology 18 (2007) 185605, https://doi.org/10.1088/0957-4484/18/18/185605. [34] A.V. Naumkin, A. Kraut-Vass, S.W. Gaarenstroom, C.J. Powell, NIST X-ray Photoelectron Spectroscopy Database, NIST Standard Reference Database 20, Version 4.1, U.S. Secretary of Commerce, (2012). https://srdata.nist.gov/xps/. [35] P.B. Amama, C.L. Pint, F. Mirri, M. Pasquali, R.H. Hauge, B. Maruyama, Catalystsupport interactions and their influence in water-assisted carbon nanotube carpet growth, Carbon 50 (2012) 2396–2406, https://doi.org/10.1016/j.carbon.2012.01. 045.

Acknowledgements We thank Michel Ramonda and the Near-Field Microscopy service (CTM) of the University of Montpellier for AFM analyses. HRTEM studies were conducted at the Laboratorio de Microscopias Avanzadas, Instituto de Nanociencia de Aragon, Universidad de Zaragoza, Spain. The Government of Aragon, and the European Social Fund are gratefully acknowledged. R.A. gratefully acknowledges the project “Construyendo Europa desde Aragon” 2014-2020 (grant number E/26). R.A. gratefully acknowledges the support from the Spanish Ministry of Economy and Competitiveness (MINECO) through project grant MAT2016-79776-P (AEI/FEDER, UE). Part of this work has received funding from the European Union’s Horizon 2020 research and innovation program under the Marie Sklodowska-Curie grant agreement No 642742. Data availability No raw/processed data are required to reproduce these findings. References [1] W.A. de Heer, A. Châtelain, D. Ugarte, A carbon nanotube field-emission electron source, Science 270 (1995) 1179–1180, https://doi.org/10.1126/science.270. 5239.1179. [2] N. Perea-López, B. Rebollo-Plata, J.A. Briones-León, A. Morelos-Gómez, D. Hernández-Cruz, G.A. Hirata, V. Meunier, A.R. Botello-Méndez, J.-C. Charlier, B. Maruyama, E. Muñoz-Sandoval, F. López-Urías, M. Terrones, H. Terrones, Millimeter-long carbon nanotubes: outstanding electron-emitting sources, ACS Nano 5 (2011) 5072–5077, https://doi.org/10.1021/nn201149y. [3] D.N. Futaba, K. Hata, T. Yamada, T. Hiraoka, Y. Hayamizu, Y. Kakudate, O. Tanaike, H. Hatori, M. Yumura, S. Iijima, Shape-engineerable and highly densely packed single-walled carbon nanotubes and their application as super-capacitor electrodes, Nat Mater. 5 (2006) 987–994, https://doi.org/10.1038/nmat1782. [4] K. Byungwoo, C. Haegeun, K. Woong, High-performance supercapacitors based on vertically aligned carbon nanotubes and nonaqueous electrolytes, Nanotechnology 23 (2012) 155401, , https://doi.org/10.1088/0957-4484/23/15/155401. [5] Y. Wang, J.T.W. Yeow, A review of carbon nanotubes-based gas sensors, J. Sens. 2009 (2009) 24, https://doi.org/10.1155/2009/493904. [6] D.R. Kauffman, A. Star, Carbon nanotube gas and vapor sensors, Angew Chem Int Ed. 47 (2008) 6550–6570, https://doi.org/10.1002/anie.200704488. [7] B.J. Hinds, N. Chopra, T. Rantell, R. Andrews, V. Gavalas, L.G. Bachas, Aligned multiwalled carbon nanotube membranes, Science 303 (2004) 62–65, https://doi. org/10.1126/science.1092048. [8] F. Du, L. Qu, Z. Xia, L. Feng, L. Dai, Membranes of vertically aligned superlong carbon nanotubes, Langmuir 27 (2011) 8437–8443, https://doi.org/10.1021/ la200995r. [9] H. Cebeci, R.G.d. Villoria, A.J. Hart, B.L. Wardle, Multifunctional properties of high volume fraction aligned carbon nanotube polymer composites with controlled morphology, Compos. Sci. Technol. 69 (2009) 2649–2656, https://doi.org/10. 1016/j.compscitech.2009.08.006. [10] V.P. Veedu, A. Cao, X. Li, K. Ma, C. Soldano, S. Kar, P.M. Ajayan, M.N. GhasemiNejhad, Multifunctional composites using reinforced laminae with carbon-nanotube forests, Nat. Mater. 5 (2006) 457–462, https://doi.org/10.1038/nmat1650. [11] T. Tong, Y. Zhao, L. Delzeit, A. Kashani, M. Meyyappan, A. Majumdar, Dense vertically aligned multiwalled carbon nanotube arrays as thermal interface materials, IEEE Trans. Compon. Packag. Technol. 30 (2007) 92, https://doi.org/10.1109/ TCAPT.2007.892079.

45

Materials Science & Engineering B 245 (2019) 37–46

A. Roussey, et al.

[39] Y. Zhang, G. Zou, S.K. Doorn, H. Htoon, L. Stan, M.E. Hawley, C.J. Sheehan, Y. Zhu, Q. Jia, Tailoring the morphology of carbon nanotube arrays: from spinnable forests to undulating foams, ACS Nano 3 (2009) 2157–2162, https://doi.org/10.1021/ nn9003988. [40] C.S. Choi, J.E. Mapes, E. Prince, The structure of ammonium nitrate (IV), Acta Crystallogr. B 28 (1972) 1357–1361, https://doi.org/10.1107/ S0567740872004303. [41] A. Yamamoto, Modulated structure of wustite (Fe1-xO) (three-dimensional modulation), Acta Crystallogr. B 38 (1982) 1451–1456, https://doi.org/10.1107/ S056774088200613X.

[36] A. Kaneko, K. Yamada, R. Kumahara, H. Kato, Y. Homma, Comparative study of catalytic activity of iron and cobalt for growing carbon nanotubes on alumina and silicon oxide, J. Phys. Chem. C 116 (2012) 26060–26065, https://doi.org/10.1021/ jp309232w. [37] M. Picher, H. Navas, R. Arenal, E. Quesnel, E. Anglaret, V. Jourdain, Influence of the growth conditions on the defect density of single-walled carbon nanotubes, Carbon. 50 (2012) 2407–2416, https://doi.org/10.1016/j.carbon.2012.01.055. [38] C.L. Cheung, A. Kurtz, H. Park, C.M. Lieber, Diameter-controlled synthesis of carbon nanotubes, J. Phys. Chem. B 106 (2002) 2429–2433, https://doi.org/10.1021/ jp0142278.

46